2100. Algorithmic Model for Accurate and Timely Diagnoses of Appendicitis in Children
Authors * Denotes Presenting Author
  1. Edward Florez *; University of Mississippi Medical Center
  2. Juliana Sitta; University of Mississippi Medical Center
  3. Sarah Miller; University of Mississippi Medical Center
  4. Aubrey Smyly; University of Mississippi Medical Center
  5. Elliot Varney; University of Mississippi Medical Center
  6. Seth Lirette; University of Mississippi Medical Center
  7. Candace Howard; University of Mississippi Medical Center
Appendicitis is the most common surgical emergency in children. The diagnostic workup of appendicitis in normal-weight children begins with ultrasound (US) followed by computed tomography (CT). However, the initial assessment of appendicitis in obese children is often insufficient due to limitations in the physical examination. Thus, obese patients are prone to undergo additional imaging assessment, potentially delaying management, increasing cost, and risk of complications. This study aims to determine the diagnostic accuracy of CT and US in children with appendicitis. We propose a model to predict which imaging modality (US vs. CT) would provide the most accurate diagnosis based on body habitus.

Materials and Methods:
This is a HIPPA-compliant, IRB approved, single-center retrospective study. Pediatric patients with a history of appendicitis-like symptoms admitted to the emergency room between 2015 and 2019 were enrolled (N=1111). Patients that underwent both CT and US were included in the study (N=396). Demographic and clinical information such as age, BMI, gender, Alvarado score were collected from the electronic medical record. Two independent readers performed anthropometric measurements using a Digital Imaging and Communications in Medicine viewer. The waist circumference (WC) was measured at the top of the iliac crest and the sagittal abdominal diameter (SAD) was measured at the anterior superior iliac spine level. Four readers contoured fat and muscle depots of a single CT slice at the L4-L5 intervertebral disc using a segmentation software. Linear Pearson coefficient was used to correlate WC and SAD with body composition and BMI. Inter-observer agreement was assessed by the intraclass correlation coefficient (ICC) with 95% confidence intervals and Bland Altman plots. Finally, a model was developed to determine the probability of seeing the appendix by US since it is the standard of care in patients with appendicitis.

The predictor model was based on BMI interactions (AUC=0.65, p<0.001). The model showed that the appendix in patients with BMI=20kg/m2 has <50% probability of being detected by US, which resulted in a high number of unidentified appendixes (>60%). This limitation was evidenced when ~30% of the diagnoses obtained using CT and US were discordant. Thus, our model suggests using CT images prior to diagnosing appendicitis or normal appendix. Based on abdominal CT images, WC and SAD demonstrated a strong correlation with BMI (R2=0.88 and R2=0.86 respectively, p<0.001) and total adipose tissue (R2=0.80 and R2=0.79 respectively, p<0.001), including subcutaneous adipose tissue (R2=0.79 and R2=0.78 respectively, p<0.001) and visceral adipose tissue (R2=0.84 each, p<0.001). Both intra- and inter-observer agreement were excellent (0.99 ICC, 95%CI 0.98–1.0).

Our model allows a rapid and easy determination of the best imaging modality to diagnose appendicitis based on body habitus while reducing the costs (eg. delay in diagnosis and misdiagnosis) inherent to US obstacles as the first-line assessment of suspected appendicitis in a diverse pediatric population.